Semantic segmentation of gonio-photographs via adaptive ROI localisation and uncertainty estimation

ObjectiveTo develop and test a deep learning (DL) model for semantic segmentation of anatomical layers of the anterior chamber angle (ACA) in digital gonio-photographs.Methods and analysisWe used a pilot dataset of 274 ACA sector images, annotated by expert ophthalmologists to delineate five anatomical layers: iris root, ciliary body band, scleral spur, trabecular meshwork and cornea. Narrow depth-of-field and peripheral vignetting prevented clinicians from annotating part of each image with sufficient confidence, introducing a degree of subjectivity and features correlation in the ground truth. To overcome these limitations, we present a DL model, designed and trained to perform two tasks simultaneously: (1) maximise the segmentation accuracy within the annotated region of each frame and (2) identify a region of interest (ROI) based on local image informativeness. Moreover, our calibrated model provides results interpretability returning pixel-wise classification uncertainty through Monte Carlo dropout.ResultsThe model was trained and validated in a 5-fold cross-validation experiment on ~90% of available data, achieving ~91% average segmentation accuracy within the annotated part of each ground truth image of the hold-out test set. An appropriate ROI was successfully identified in all test frames. The uncertainty estimation module located correctly inaccuracies and errors of segmentation outputs.ConclusionThe proposed model improves the only previously published work on gonio-photographs segmentation and may be a valid support for the automatic processing of these images to evaluate local tissue morphology. Uncertainty estimation is expected to facilitate acceptance of this system in clinical settings.

Assessing Land Use and Land Cover Change in the Volta Grande Do Xingu (Pará – Brazil) Between 2000 and 2017

This study aims to assess the land use and land cover change through the use of three pixel-based methods of image classification (Mahalanobis, Maximum Likelihood, and Minimum Distance) in the region of Volta Grande do Xingu (Brazilian Amazon), under influence of the Belo Monte hydroelectric power plant. Different pixel-based classification methods were performed on Landsat 7 and 8 multispectral products from the years 2000 and 2017, using a 2008 map as the ground truth image. The accuracies of the classifications were compared, and land use change analyses were performed in the different scenarios. The main impacts regarding land use and land cover change were from forest to agro pasture, from non-river to river upstream the Xingu river, and from river to non-river in the south of the Volta Grande do Xingu, resulting in rocks exposure.

Evaluating the Coding Performance of 360° Image Projection Formats Using Objective Quality Metrics

Recently, 360° content has emerged as a new method for offering real-life interaction. Ultra-high resolution 360° content is mapped to the two-dimensional plane to adjust to the input of existing generic coding standards for transmission. Many formats have been proposed, and tremendous work is being done to investigate 360° videos in the Joint Video Exploration Team using projection-based coding. However, the standardization activities for quality assessment of 360° images are limited. In this study, we evaluate the coding performance of various projection formats, including recently-proposed formats adapting to the input of JPEG and JPEG 2000 content. We present an overview of the nine state-of-the-art formats considered in the evaluation. We also propose an evaluation framework for reducing the bias toward the native equi-rectangular (ERP) format. We consider the downsampled ERP image as the ground truth image. Firstly, format conversions are applied to the ERP image. Secondly, each converted image is subjected to the JPEG and JPEG 2000 image coding standards, then decoded and converted back to the downsampled ERP to find the coding gain of each format. The quality metrics designed for 360° content and conventional 2D metrics have been used for both end-to-end distortion measurement and codec level, in two subsampling modes, i.e., YUV (4:2:0 and 4:4:4). Our evaluation results prove that the hybrid equi-angular format and equatorial cylindrical format achieve better coding performance among the compared formats. Our work presents evidence to find the coding gain of these formats over ERP, which is useful for identifying the best image format for a future standard.

Synthetic aperture radar image segmentation using supervised artificial neural network

Synthetic Aperture Radar Image Segmentation has been a challenging task because of the presence of speckle noise. Therefore, the segmentation process can not directly rely on the intensity information alone, but it must consider several derived features in order to get satisfactory segmentation results. In this paper, it is attempted to use supervised information about regions for segmentation criteria in which ANN is employed to give training on the basis of known ground truth image derived. Three different features are employed for segmentation, first feature is the original image, second feature is the roughness information and the third feature is the filtered image. The segmentation accuracy is measured against the Difficulty of Segmentation (DoS) and Cross Region Fitting (CRF) methods. The performance of our algorithm has been compared with other proposed methods employing the same set of data.

GENERATIVE ADVERSARIAL NETWORKS AS A NOVEL APPROACH FOR TECTONIC FAULT AND FRACTURE EXTRACTION IN HIGH-RESOLUTION SATELLITE AND AIRBORNE OPTICAL IMAGES

We develop a novel method based on Deep Convolutional Networks (DCN) to automate the identification and mapping of fracture and fault traces in optical images. The method employs two DCNs in a two players game: a first network, called Generator, learns to segment images to make them resembling the ground truth; a second network, called Discriminator, measures the differences between the ground truth image and each segmented image and sends its score feedback to the Generator; based on these scores, the Generator improves its segmentation progressively. As we condition both networks to the ground truth images, the method is called Conditional Generative Adversarial Network (CGAN). We propose a new loss function for both the Generator and the Discriminator networks, to improve their accuracy. Using two criteria and a manually annotated optical image, we compare the generalization performance of the proposed method to that of a classical DCN architecture, U-net. The comparison demonstrates the suitability of the proposed CGAN architecture. Further work is however needed to improve its efficiency.

Segmentation and Evaluation of COVID-19 Lesion from CT scan Slices - A Study with Kapur/Otsu Function and Cuckoo Search Algorithm

Infection/disease in lung is one of the acute illnesses in humans. Pneumonia is one of the major lung diseases and each year; the death rate due to the untreated pneumonia is on rise globally. From December 2019; the pneumonia caused by the Coronavirus Disease (COVID-19) has emerged as a global threat due to its rapidity. The clinical level assessment of the COVID-19 is normally performed with the Computed-Tomography scan Slice (CTS) or the Chest X-ray. This research aims to propose an image processing system to examine the COVID-19 infection in CTS. This work implements Cuckoo-Search-Algorithm (CSA) monitored Kapur/Otsu image thresholding and a chosen image segmentation procedure to extract the pneumonia infection. After extracting the COVID-19 infection from the CTS, a relative assessment is then executed with the Ground-Truth-Image (GTI) offered by a radiologist and the essential performance measures are then computed to confirm the superiority of the proposed technique. This work also presents a comparative assessment among the segmentation procedures, such as Level-Set (LS) and Chan-Vese (CV) methods. The experimental outcome authenticates that, the results by Kapur and Otsu threshold are approximately similar when the LS is implemented and the CV with the Otsu presents better values of Jaccard, Dice and Accuracy compared to other methods presented in this research.

Mathematical Modelling of Ground Truth Image for 3D Microscopic Objects Using Cascade of Convolutional Neural Networks Optimized with Parameters’ Combinations Generators

Mathematical modelling to compute ground truth from 3D images is an area of research that can strongly benefit from machine learning methods. Deep neural networks (DNNs) are state-of-the-art methods design for solving these kinds of difficulties. Convolutional neural networks (CNNs), as one class of DNNs, can overcome special requirements of quantitative analysis especially when image segmentation is needed. This article presents a system that uses a cascade of CNNs with symmetric blocks of layers in chain, dedicated to 3D image segmentation from microscopic images of 3D nuclei. The system is designed through eight experiments that differ in following aspects: number of training slices and 3D samples for training, usage of pre-trained CNNs and number of slices and 3D samples for validation. CNNs parameters are optimized using linear, brute force, and random combinatorics, followed by voter and median operations. Data augmentation techniques such as reflection, translation and rotation are used in order to produce sufficient training set for CNNs. Optimal CNN parameters are reached by defining 11 standard and two proposed metrics. Finally, benchmarking demonstrates that CNNs improve segmentation accuracy, reliability and increased annotation accuracy, confirming the relevance of CNNs to generate high-throughput mathematical ground truth 3D images.

Diversity-Generated Image Inpainting with Style Extraction

The latest methods based on deep learning have achieved amazing results regarding the complex work of inpainting large missing areas in an image. This type of method generally attempts to generate one single "optimal" inpainting result, ignoring many other plausible results. However, considering the uncertainty of the inpainting task, one sole result can hardly be regarded as a desired regeneration of the missing area. In view of this weakness, which is related to the design of the previous algorithms, we propose a novel deep generative model equipped with a brand new style extractor which can extract the style noise (a latent vector) from the ground truth image. Once obtained, the extracted style noise and the ground truth image are both input into the generator. We also craft a consistency loss that guides the generated image to approximate the ground truth. Meanwhile, the same extractor captures the style noise from the generated image, which is forced to approach the input noise according to the consistency loss. After iterations, our generator is able to learn the styles corresponding to multiple sets of noise. The proposed model can generate a (sufficiently large) number of inpainting results consistent with the context semantics of the image. Moreover, we check the effectiveness of our model on three databases, i.e., CelebA, Agricultural Disease, and MauFlex. Compared to state-of-the-art inpainting methods, this model is able to offer desirable inpainting results with both a better quality and higher diversity. The code and model will be made available on https://github.com/vivitsai/SEGAN.

Asymmetric Encoder-Decoder Structured FCN Based LiDAR to Color Image Generation

In this paper, we propose a method of generating a color image from light detection and ranging (LiDAR) 3D reflection intensity. The proposed method is composed of two steps: projection of LiDAR 3D reflection intensity into 2D intensity, and color image generation from the projected intensity by using a fully convolutional network (FCN). The color image should be generated from a very sparse projected intensity image. For this reason, the FCN is designed to have an asymmetric network structure, i.e., the layer depth of the decoder in the FCN is deeper than that of the encoder. The well-known KITTI dataset for various scenarios is used for the proposed FCN training and performance evaluation. Performance of the asymmetric network structures are empirically analyzed for various depth combinations for the encoder and decoder. Through simulations, it is shown that the proposed method generates fairly good visual quality of images while maintaining almost the same color as the ground truth image. Moreover, the proposed FCN has much higher performance than conventional interpolation methods and generative adversarial network based Pix2Pix. One interesting result is that the proposed FCN produces shadow-free and daylight color images. This result is caused by the fact that the LiDAR sensor data is produced by the light reflection and is, therefore, not affected by sunlight and shadow.